Plasticity of perisynaptic astroglia during ischemia-induced spreading depolarization

Abstract

High astroglial capacity for glutamate and potassium clearance aids in recovering spreading depolarization (SD)-evoked disturbance of ion homeostasis during stroke. Since perisynaptic astroglia cannot be imaged with diffraction-limited light microscopy, nothing is known about the impact of SD on the ultrastructure of a tripartite synapse. We used serial section electron microscopy to assess astroglial synaptic coverage in the sensorimotor cortex of urethane-anesthetized male and female mice during and after SD evoked by transient bilateral common carotid artery occlusion. At the subcellular level, astroglial mitochondria were remarkably resilient to SD compared to dendritic mitochondria that were fragmented by SD. Overall, 482 synapses in `Sham' during `SD' and `Recovery' groups were randomly selected and analyzed in 3D. Perisynaptic astroglia was present at the axon-spine interface (ASI) during SD and after recovery. Astrocytic processes were more likely found at large synapses on mushroom spines after recovery, while the length of the ASI perimeter surrounded by astroglia has also significantly increased at large synapses. These findings suggest that as larger synapses have a bigger capacity for neurotransmitter release during SD, they attract astroglial processes to their perimeter during recovery, limiting extrasynaptic glutamate escape and further enhancing the astrocytic ability to protect synapses in stroke.

Keywords: astrocyte; mitochondria; quantitative serial section electron microscopy; stroke; tripartite synapse.

Fig. 1

Experimental design to examine the impact of a single SD on the cortical ultrastructure. A) Spontaneous ECoG activity (top trace, bandpass filtered at 1–3 Hz) and corresponding DC cortical surface potential (bottom trace) recorded by a glass microelectrode placed in layer I of the mouse sensorimotor cortex. Rapid ECoG suppression during BCCA ligation was followed by a single SD represented by an abrupt negative shift of the DC potential. Both ECoG and the DC potential recovered after reperfusion. Green arrows specify time points corresponding to the start of occlusion and the beginning of blood reperfusion at ~15 s after SD onset. Blue arrows indicate approximate time points on the DC trace when mice in Sham and Recovery groups were perfusion-fixed with aldehydes through the heart. The red arrow indicates that mice in the SD group were perfusion-fixed during depolarization without blood supply returning to the cortex. B1, C1, D1) Color-coded representative electron micrographs from layer I of the sensorimotor cortex at the location of electrophysiological recordings from Sham, SD, and Recovery groups. Astrocytic processes (as) are illustrated in blue, dendrites (d) and spines (sp) in yellow, axons (ax) in green, and postsynaptic densities (psd) in red. The identity of astrocytic processes on these single EM sections was verified by inspecting serial sections. B2, C2, D2) Examples of spongious 3D architectonics of astroglia reconstructed through serial sections in Sham, SD, and Recovery groups.

Fig. 2

Astroglial mitochondria were unaffected by SD, while dendritic mitochondria became swollen and fragmented. A) Representative 3D mitochondrial reconstructions from ssEM reveal that astroglial mitochondria structure was stable during SD. B) Examples of reconstructed dendritic mitochondria reveal that Sham mice contained mostly filamentous (tubular) mitochondria. In contrast, dendritic mitochondria in mice perfusion-fixed during SD were mainly in the form of swollen “mitochondria-on-a-string” (MOAS) and swollen globular mitochondria. Mainly tubular dendritic mitochondria were found again in the Recovery condition. C) Box and whisker plot showing no significant difference in the maximum diameter of astroglial mitochondria (P = 0.84 for Sham vs. SD groups; P = 0.2 for Sham vs. Recovery, and P = 0.24 for SD vs. Recovery groups; t-test in LMM). Circles depict maximum outliers. Numbers in parentheses indicate the number of mitochondria included in the analyses. D) Quantification of the maximum diameter of dendritic mitochondria showed that mitochondria were swollen in the SD group (P < 0.0001 for SD vs. Sham groups; P = 0.01 for SD vs. Recovery groups; P = 0.29 for Sham vs. Recovery; t-test in LMM). Circles depict maximum outliers.

Fig. 3

The size of docked vesicle pool decreased during SD but recovered with repolarization during blood reperfusion. A) Representative electron micrograph shows docked vesicle (arrow) adjacent to the plasma membrane of the presynaptic axonal bouton associated with a thin spine. B–D) Examples of 3D reconstructions of dendritic spines (sp, yellow) and axonal boutons (ax, green) illustrate docked vesicles (purple) next to postsynaptic densities (psd, red) in Sham (B), SD (C), and Recovery (D) groups. E) Regression lines show a positive correlation between docked vesicles number and PSD area in the Sham (R = 0.57, P < 0.0001), SD (R = 0.58, P < 0.0001), and Recovery (R = 0.67, P < 0.0001) groups. The slopes of the regression line differ significantly between SD and Sham (P < 0.001; t-test in LMM) and SD and Recovery groups (P < 0.0001; t-test in LMM), with no significant difference between Sham and Recovery datasets (P = 0.13; t-test in LMM). F) Cumulative histogram of docked vesicle densities (number of vesicles/μm² of PSD area). The density was significantly lower in the SD than in the Sham (P < 0.0001, K–S test) and the Recovery (P < 0.0001, K–S test) groups, with no significant differences between docked vesicle densities in Sham and Recovery datasets (P = 0.25, K-S test). G) Docked vesicle number has significantly declined during SD on both larger synapses at mushroom spines and smaller synapses at thin spines (Mann–Whitney rank-sum test). Numbers in parentheses indicate the number of spines included in the analyses.

Fig. 4

The astrocytic content and percent of synapses with astroglia were not different between conditions. A) The astrocytic volume fraction was similar between conditions (H₍₂₎ = 0.115, P = 0.97, Kruskal–Wallis one-way ANOVA on ranks). B) 3D reconstructions illustrate locations of perisynaptic astroglia. Left: Astroglial process (as, turquoise) is positioned at the edge of the synapse (psd, red) at the ASI between the dendritic spine (sp, yellow) and presynaptic axonal bouton (ax, green) membranes. The Middle, astroglial process is located at the postsynaptic site, hovering only around the dendritic spine. Right: The astroglial process is located at the presynaptic site only next to the periphery of the presynaptic axonal bouton. C) The distribution of perisynaptic astroglia among synapses was not different between conditions (χ²₍₈₎ = 8.7, P = 0.37, χ² test).

Fig. 5

Relationship between synapse size and perisynaptic astroglia at ASI. A) Synapses were significantly larger when perisynaptic astroglia was present at ASI. Box and whisker plot showing significant PSD size differences between synapses with and without astroglia at ASI (LMM). Numbers in parentheses indicate the number of synapses included in the analyses. Circles depict maximum outliers. B) Cumulative histogram of PSD area of synapses with astroglial processes apposing ASI and without astroglia at ASI. Synapses with astroglia at ASI had significantly bigger PSDs (P < 0.005, K–S test). C) Synapses are larger on mushrooms than on thin spines in all conditions (P < 0.0001; LMM). The PSD area of mushroom and thin spines remains unchanged across experimental groups (NS; t-test in LMM). Numbers in parentheses indicate the number of synapses included in the analyses. D) 237 mushroom spines, 209 thin spines, and 34 “other” synapses were analyzed across 3 experimental groups. The fraction of larger synapses with perisynaptic astroglia has increased after recovery from SD. There was an increase in the percentage of mushroom spines with astroglia at ASI (χ²₍₂₎ = 8.05, P = 0.02, χ² test), while there were no changes in the percentage of thin spines (χ²₍₂₎ = 0.09, P = 0.96, χ² test) and “other” synapses (χ²₍₂₎ = 0.36, P = 0.84, χ² test) with astrocytic processes at ASI across conditions.

Fig. 6

The length of the synaptic perimeter surrounded by perisynaptic astroglia has significantly increased at larger synapses. A) Reconstructed dendritic segment (den, yellow) with a mushroom spine (sp, yellow), postsynaptic density (psd, red), axonal bouton (ax, green), and perisynaptic astroglial process (as, turquoise) adjacent to the edge of ASI. Apposed by astroglia ASI is outlined by the purple line, with the rest of the astroglial-free ASI perimeter remaining unmarked (green arrow). The astroglial process apposed approximately 45% of the ASI of this mushroom spine. B) Regression analyses show a positive correlation between the length of the apposed ASI and the PSD size in the Sham (R = 0.32, P < 0.0001; n = 160 synapses), SD (R = 0.48, P < 0.0001; n = 160 synapses), and Recovery (R = 0.59, P < 0.0001; n = 161 synapses) groups. The slopes of the regression lines differ significantly between Sham and Recovery (P < 0.00001; t-test in LMM) and SD and Recovery groups (P = 0.0043; t-test in LMM), with no significant difference between Sham and SD datasets (P = 0.05; t-test in LMM), indicating an increase in the length of the synaptic perimeter with astroglia in the Recovery dataset. C) Regression lines show a positive correlation between the length of apposed ASI at large synapses on mushroom spines and the PSD size in the Sham (R = 0.29, P = 0.009; n = 79 spines), SD (R = 0.37, P < 0.001; n = 79), and Recovery groups (R = 0.52, P < 0.0001; n = 78 spines). There was a significant difference in the slopes of regression lines between Sham and Recovery (P = 0.002; t-test in LMM) and SD and Recovery groups (P = 0.013; t-test in LMM), with no significant difference between Sham and SD mice (P = 0.46; t-test in LMM), indicating that perisynaptic astroglial ensheathment has increased at the mushroom spines in the Recovery group. D) Regression lines show a positive correlation between the length of apposed ASI at synapses on thin spines and the PSD area in the Sham (R = 0.29, P = 0.012; n = 71 spines), SD (R = 0.57, P < 0.0001; n = 68), and Recovery groups (R = 0.45, P < 0.001; n = 70 spines). The slopes of regression lines were not different between groups (P = 0.68, Sham vs. Recovery; P = 0.1, SD vs. Recovery; P = 0.06, Sham vs. SD; t-test in LMM), indicating that astroglial coverage at ASI was not different at thin spines.

Fig. 7

Summary of the study. Astroglial mitochondria are remarkably resilient to a short-lasting ischemia-induced SD compared to dendritic mitochondria. Larger synapses with a bigger capacity for glutamate release are associated with increased astrocytic apposition after recovery from short-lasting ischemia-induced SD, limiting extrasynaptic glutamate spillover and further enhancing the astrocytic ability to protect synaptic circuitry in stroke.